| 1 | //! Importing a repository from another git host. |
| 2 | //! |
| 3 | //! This finds a public repository over smart HTTP and its default branch, so |
| 4 | //! an address that does not work is refused before anything is made; the |
| 5 | //! copy itself, every branch and tag, is mirror.rs's (`Endpoint::anonymous`). |
| 6 | //! Only public repositories reachable over https are imported this way. |
| 7 | |
| 8 | use worker::js_sys::Uint8Array; |
| 9 | use worker::{Fetch, Headers, Method, Request, RequestInit, Result, Url}; |
| 10 | |
| 11 | use crate::land::read_pkt_lines; |
| 12 | |
| 13 | const HEADS: &str = "refs/heads/"; |
| 14 | /// Some hosts only speak the smart protocol to something that says it is git. |
| 15 | const USER_AGENT: &str = "git/2.45.0 (g1t import)"; |
| 16 | |
| 17 | /// What the other host says its default branch is and where it points. |
| 18 | #[derive(Debug, PartialEq, Eq)] |
| 19 | pub struct Remote { |
| 20 | pub branch: String, |
| 21 | pub head: String, |
| 22 | } |
| 23 | |
| 24 | /// The address to import from, tidied, or `None` if it is not one g1t will |
| 25 | /// fetch: it must be https, with no credentials in it. |
| 26 | pub fn clean_url(url: &str) -> Option<String> { |
| 27 | let parsed = Url::parse(url.trim()).ok()?; |
| 28 | let plain = parsed.scheme() == "https" |
| 29 | && parsed.username().is_empty() |
| 30 | && parsed.password().is_none() |
| 31 | && parsed.host_str().is_some() |
| 32 | && parsed.query().is_none(); |
| 33 | if !plain { |
| 34 | return None; |
| 35 | } |
| 36 | let host = parsed.host_str()?; |
| 37 | let path = parsed.path().trim_end_matches('/'); |
| 38 | (path.len() > 1).then(|| format!("https://{host}{path}")) |
| 39 | } |
| 40 | |
| 41 | /// The default branch and its head, from a ref advertisement. |
| 42 | fn parse_remote(bytes: &[u8]) -> Option<Remote> { |
| 43 | let (lines, _) = read_pkt_lines(bytes); |
| 44 | let mut head = None; |
| 45 | let mut branch = None; |
| 46 | let mut branches = Vec::new(); |
| 47 | for line in lines { |
| 48 | let mut parts = line.splitn(2, |byte| *byte == 0); |
| 49 | let reference = std::str::from_utf8(parts.next()?).ok()?.trim_end(); |
| 50 | // The first ref carries the capabilities, one of which names the |
| 51 | // branch HEAD points to. |
| 52 | if let Some(capabilities) = parts |
| 53 | .next() |
| 54 | .and_then(|bytes| std::str::from_utf8(bytes).ok()) |
| 55 | { |
| 56 | branch = capabilities |
| 57 | .split(' ') |
| 58 | .find_map(|capability| capability.trim().strip_prefix("symref=HEAD:refs/heads/")) |
| 59 | .map(str::to_owned); |
| 60 | } |
| 61 | let Some((hash, name)) = reference.split_once(' ') else { |
| 62 | continue; |
| 63 | }; |
| 64 | if name == "HEAD" { |
| 65 | head = Some(hash.to_owned()); |
| 66 | } else if let Some(name) = name.strip_prefix(HEADS) { |
| 67 | branches.push((name.to_owned(), hash.to_owned())); |
| 68 | } |
| 69 | } |
| 70 | // Without a symref, the branch HEAD agrees with; failing that, main. |
| 71 | let branch = branch.or_else(|| { |
| 72 | let head = head.as_deref()?; |
| 73 | branches |
| 74 | .iter() |
| 75 | .find(|(_, hash)| hash == head) |
| 76 | .map(|(name, _)| name.clone()) |
| 77 | })?; |
| 78 | let head = branches |
| 79 | .iter() |
| 80 | .find(|(name, _)| *name == branch) |
| 81 | .map(|(_, hash)| hash.clone()) |
| 82 | .or(head)?; |
| 83 | Some(Remote { branch, head }) |
| 84 | } |
| 85 | |
| 86 | fn request(method: Method, url: &str, body: Option<Vec<u8>>) -> Result<Request> { |
| 87 | let headers = Headers::new(); |
| 88 | headers.set("user-agent", USER_AGENT)?; |
| 89 | if body.is_some() { |
| 90 | headers.set("content-type", "application/x-git-upload-pack-request")?; |
| 91 | headers.set("accept", "application/x-git-upload-pack-result")?; |
| 92 | } |
| 93 | let mut init = RequestInit::new(); |
| 94 | init.with_method(method).with_headers(headers); |
| 95 | if let Some(body) = body { |
| 96 | init.with_body(Some(Uint8Array::from(body.as_slice()).into())); |
| 97 | } |
| 98 | Request::new_with_init(url, &init) |
| 99 | } |
| 100 | |
| 101 | /// Asks the other host what it has. `Err` in the inner result is a reason |
| 102 | /// to show the person importing. |
| 103 | pub async fn discover(url: &str) -> Result<std::result::Result<Remote, String>> { |
| 104 | let request = request( |
| 105 | Method::Get, |
| 106 | &format!("{url}/info/refs?service=git-upload-pack"), |
| 107 | None, |
| 108 | )?; |
| 109 | let mut response = match Fetch::Request(request).send().await { |
| 110 | Ok(response) => response, |
| 111 | Err(_) => return Ok(Err("That address could not be reached.".to_owned())), |
| 112 | }; |
| 113 | if response.status_code() != 200 { |
| 114 | return Ok(Err( |
| 115 | "No public git repository was found at that address. Private repositories cannot be imported." |
| 116 | .to_owned(), |
| 117 | )); |
| 118 | } |
| 119 | let bytes = response.bytes().await?; |
| 120 | Ok(parse_remote(&bytes).ok_or_else(|| "That repository is empty.".to_owned())) |
| 121 | } |
| 122 | |
| 123 | #[cfg(test)] |
| 124 | mod tests { |
| 125 | use super::*; |
| 126 | |
| 127 | fn pkt(payload: &str) -> Vec<u8> { |
| 128 | format!("{:04x}{payload}", payload.len() + 4).into_bytes() |
| 129 | } |
| 130 | |
| 131 | #[test] |
| 132 | fn only_plain_https_addresses_are_fetched() { |
| 133 | assert_eq!( |
| 134 | clean_url(" https://github.com/syntaqx/hello/ ").as_deref(), |
| 135 | Some("https://github.com/syntaqx/hello") |
| 136 | ); |
| 137 | assert_eq!( |
| 138 | clean_url("https://github.com/syntaqx/hello.git").as_deref(), |
| 139 | Some("https://github.com/syntaqx/hello.git") |
| 140 | ); |
| 141 | for bad in [ |
| 142 | "http://github.com/a/b", |
| 143 | "git@github.com:a/b.git", |
| 144 | "https://user:secret@github.com/a/b", |
| 145 | "https://github.com", |
| 146 | "https://github.com/a/b?x=1", |
| 147 | "not a url", |
| 148 | ] { |
| 149 | assert_eq!(clean_url(bad), None, "{bad}"); |
| 150 | } |
| 151 | } |
| 152 | |
| 153 | #[test] |
| 154 | fn the_default_branch_comes_from_the_symref() { |
| 155 | let trunk = "c71546fcd893ef8b0f57388b65e620d759705dda"; |
| 156 | let other = "4807077b296e6edbf410d55e72749d3e1170c291"; |
| 157 | let advertisement = [ |
| 158 | pkt("# service=git-upload-pack\n"), |
| 159 | b"0000".to_vec(), |
| 160 | pkt(&format!( |
| 161 | "{trunk} HEAD\0multi_ack side-band-64k symref=HEAD:refs/heads/trunk agent=git/x\n" |
| 162 | )), |
| 163 | pkt(&format!("{other} refs/heads/feature\n")), |
| 164 | pkt(&format!("{trunk} refs/heads/trunk\n")), |
| 165 | b"0000".to_vec(), |
| 166 | ] |
| 167 | .concat(); |
| 168 | assert_eq!( |
| 169 | parse_remote(&advertisement), |
| 170 | Some(Remote { |
| 171 | branch: "trunk".to_owned(), |
| 172 | head: trunk.to_owned() |
| 173 | }) |
| 174 | ); |
| 175 | } |
| 176 | |
| 177 | #[test] |
| 178 | fn without_a_symref_the_branch_head_points_to_is_used() { |
| 179 | let main = "c71546fcd893ef8b0f57388b65e620d759705dda"; |
| 180 | let advertisement = [ |
| 181 | pkt("# service=git-upload-pack\n"), |
| 182 | b"0000".to_vec(), |
| 183 | pkt(&format!("{main} HEAD\0side-band-64k\n")), |
| 184 | pkt(&format!("{main} refs/heads/main\n")), |
| 185 | b"0000".to_vec(), |
| 186 | ] |
| 187 | .concat(); |
| 188 | assert_eq!(parse_remote(&advertisement).unwrap().branch, "main"); |
| 189 | let empty = [pkt("# service=git-upload-pack\n"), b"00000000".to_vec()].concat(); |
| 190 | assert_eq!(parse_remote(&empty), None); |
| 191 | } |
| 192 | } |